This episode is the Section 112 in Technology panel from the Engelberg Center's Disclosure Under Section 112 in Policy and Practice event. It was recorded on April 19, 2023.
Laura Sheridan (Moderator), Google
Isabella Fu, Microsoft
David J. Kappos, Cravath, Swaine & Moore LLP
Natalie S. Richer, Lerner David
Announcer 0:00
Welcome to Engelberg Center Live!, a collection of audio from events held by the Engelberg Center on Innovation Law and Policy at NYU Law. This episode is the Section 112 in Technology panel from the Engelberg Center's Disclosure Under Section 112 in Policy and Practice event. It was recorded on April 19 2023.
Michael Weinberg 0:25
All right, thank you, everyone, we are going to get started with the secondary last panel. The last panel was all about how 112 applies in the life sciences. And we wanted to take the opportunity this afternoon to kind of break down a little bit more specific in the industry focus industries that are focused on how these questions impact them specifically. So the panel we're going to move into right now is all about how 112 impact in the tech industry. And so I will hand it over to Laura to get a roll.
Laura Sheridan 0:56
Great. Thank you so much, Michael. And thanks to NYU for for hosting this timely event. It has been really, really awesome so far. So we're going to shift gears, as Michael said, to talk about the tech industry, I think we've had a very life scientists focus day, you know, talking about Amgen, and what is going on in that case. And so our role is really to spell some of this out for tech specifically. And see, I think how some of the issues will differ from what we've been talking about throughout the day so far. So unlike what I think we heard from the last panel, and what Henry was talking about, I think in tech on the whole written description and enablement are viewed as relatively low hurdles. They're not they're not really constraining claim breadth in the way that they might be doing. So in the life sciences field. And I think we're seeing is artificial intelligence is growing in usage. There are going to be difficult questions coming up on enablement and written description as well. So maybe that that issue will come to head for tech. But as of now, that's really been a relatively low hurdle to face. What I do think you hear a lot more about and the the previous panel really teed it up nicely, is this issue of functional claiming, and specifically, when section 112 f should govern claim interpretation? That question keeps coming up all day, given the nature of the claims in the Amgen case. And that is really where our panel is ultimately going to end up. So we will talk to John's questions and bring in some of these themes about functional claiming and and speak to you to Bruce asking, should we be doing this at all, we'll be talking about that. Of course, that will bring in the issue of definiteness, which I think hasn't been covered all that much today, either in Section 112 B. So we have an amazing panel here to talk about these topics, we have Isabella foo down at the end, who's head of patents at Microsoft, she was previously head of IP litigation. And during that time, she was instrumental on one of the biggest 112 cases, we have Williamson, so I have a feeling she'll talk about that today. We have Dave Cabos, who of course, is the former head of the USPTO. During his time at the office, he was grappling with a lot of these issues, particularly around software claims. So I think I'll speak to some of that as well. And we have Natalie richer, who is a partner at learner, David. And she works with clients in the tech industry to develop their patent portfolios. And she's an electrical engineer by training. So we have an amazing panel. And we're going to jump right in. And as much as I said, we're going to talk about functional claiming, we'll put a pin in that for the moment, because I do want to connect up the tech industry issues on enablement and written description to what we've been talking about already today. So let's start with those core issues, which which is conferences covered well, and I'm going to turn to Dave. Dave, if you could give us some background on the law of enablement, written description, as applied to software and tech claims, and your assessment of how courts are doing.
David J. Kappos 4:01
Yeah, okay. Well, thanks, Laura, and great to be here, great to see friends at NYU Law School, and to be talking about such an critical topic, and one that in the past has not gotten enough coverage. And in part, thanks, I think to the injured case, which we can't get away from getting a lot of coverage. So just to set the scene a little bit. Start with the very basics. Of course, 112 specifies in section a subsection A, that the specification has to contain a written description, right, that has got to enable a person of skill in the art to to make the invention there's still the words best mode in there, but they don't really matter anymore. And so in recent years, there's been cases in the farm area but we won't really focus on that so much. I'm thinking of the area Lilly case, and others that have done a lot to tighten up. Section 112. But in the in more in the software area, I think you have to start the discussion with the ones factors. And, you know, they were where the Federal Circuit finally came in, and gave us a lot of detail back in 1988, so a while ago on, you know how to interpret 112. And this set of a different criteria, and I won't read them all, but it's, you know, the nature of the invention, the nature of the prior art, the breadth of the claims, those kinds of things in interpreting to help interpret claims under 112. Then beyond that, the best, you know, sort of introduction that back to Laura's question that I can help with, in general. And then I want to talk about software for just a second before stopping here is, you know, cases where the Federal Circuit has tried to cabin 112 and claims under 112 in various ways, such as declaring, right that the specification can always support and expand expensive claim language and satisfy the requirements of 112 merely by clearly describing one embodiment of the claimed invention. That was the lizard tech case. And it starts to give you a sense of the Federal Circuit, you know, grappling in, you know, in not a life sciences context, with the genus species issue, and the tension between, right and enablement that, and a written description that is clear on one embodiment or maybe a few embodiment, but then the desire to claim claim well beyond that, and there's been a series of cases. Another one that comes to mind, for me is the DreamWorks case, where the Federal Circuit was looking at Broad claims. I think this is, before that time of Judge Jen on the Federal Circuit, looking at, you know, broad claims relating to integrating audio signals and video images, and grappling with where the full scope of the claims isn't covered by the disclosure. But only certain embodiments are covered in the video game settings that are provided in that particular claim and that particular specification. And the court again, binding itself, not wanting to provide super broad like genius level claims go for claims that weren't enabled that broadly. So whereas life science is very different in some ways. You know, as I was looking through these cases, and preparing this panel, I found myself saying, you know, I'm reminding myself that's kind of similar in a lot of ways to in the Federal Circuit has been gradually constraining 112 shorter life scientists, but also, I think, in the tech area that we're talking about on this panel. And then lastly, I said, I'd come back to the second part of Laura's question, which was artificial Well, which was software. And here, I want to talk about some of the AI, stuff that's going on, if you will. And there, of course, have not been very many cases that have focused on AI, and the disclosure requirements. But it's becoming clear that the field is sufficiently unpredictable from the viewpoint of blackbox AI, and, and turning algorithms loose on data. And frankly, not really knowing what they're going to come out with and therefore being, you know, in many times, good way surprised by it. But I feel like there's a tension brewing in that area. An example, right, a patent issued on a PCT application that covered the use of code, an artificial neural network, having waiting values that are determined by learning. And that claim was issued from a PCT application in the US. The EPO took that same claim, and said, No, we're not going to issue that it fails to disclose which input data are suitable for training the artificial artificial intelligence neural network, according to the invention. Quoting that again, and I bring that up to say, it's an example in my mind of how well both the USPTO and EPO are struggling, but how a sort of a just a slightly different interpretive framework, you'll grant claim Seems like that at least out of the office out of the USPTO that the EPO never would have granted and then downstream, you know, that produces potentially conflicts that cause the court to once again, draw the boundaries of 112 in. And I thought that was just an interesting example of how the office has come out a little bit differently. So the bottom line, and then I'll stop is that the biotech trend, I think, is applicable to software. We're trending toward a narrower view of 112, including in the software area, and approach to 102, a, in written description and enablement that are requiring the quote, full scope of the claims right, to be enabled and not permitting anywhere near as much genius claiming based on limited species disclosure.
Speaker 3 10:55
Thanks, Dave. That's a really, really good start and a good table setting. I'll turn to Isabella then to, to talk a little bit about in practice, in your experience. How is that playing out? And Dave mentioned lizard tech? I think that's back in 2005. Has that has that had an impact? How is all this happening in litigation?
Isabella Fu 11:15
Yeah, well, I think there's um, when I think about 112, there's actually several different issues. One is the really broad claiming like the functional claim we've been talking about and the Williams in cases. One example of that I don't I don't think people usually think that that is the Microsoft case. But that was, I was the in house lawyer on that case. And it's interesting. Go here in the life sciences panel, because we were one of many, many defendants in that case. And we were the only ones who wanted to push the 112 issue back, like I remember, you know, to get to war stories, but you know, arguing with the other defendants about like, how much time I could have at Markman hearing on this issue. And it turned out to be the issue that, you know, you know, a lot of ways won the case, right. And so I don't know that the law has changed that much. In the time, you know, in the past one years, I mean, I can remember winning cases on 112, like, you know, back in 2003 2004, like there was the MIT abacus case, which was relied upon in the Williamson case, as well. And there was like lightning world. So I think it's been around for a while, but for whatever reason, it has not been that popular with a lot for people to litigate. And in many of these MIT advocates was also a case that we can we were the lead defendant on the 112 issue, and, you know, winning huge chunk of the case. But again, I had 200 defendants, then also sort of had an argument about the Markman, hearing how much time we would spend on that. So for whatever reason, and it's interesting, because I hear on the life sciences as well, people, it may be has something to do with the first panel about people feeling like it's a really hard issue to win. But in my experience, it's not necessarily it really depends upon the facts of your case. But, so we're gonna get to one drop off a little bit more. But I also want to sort of take a step back and say, I think in the computer science field, it's not just about breadth of claims, or whether the claims are valid or invalid, it's really about claim construction in the claims scope. And so when I think about what I think, written description is really for it is to tether, as we've discussed before it is to tether the claims to what was actually invented and what was disclosed. And there are many situations, we have where a patent is filed on a specific invention. So like, I think, you know, there's this case Ruckus Wireless from Federal Circuit in 2017, where the the issue was what, you know, the term communications path meant. And so what was disclosed was a wired communications path. And the disputed claims question was whether or not it could also support a wireless path. And so, you know, it was decided that no, the only thing was disclosed, was wired and wireless presented different challenges. Right. So that's an example of, you know, it's not necessarily the whole claim is functional abroad, but like, what was disclosed and what what's actually covered by the claim. I mean, another set of circumstances that we face, our selves that we see a lot is situations where, you know, we had another case, the RCT case. And that dealt with sort of halftone the technology where the display or the printer can, can render more, you know, more shades, you know, with with a fewer amount of colors by, you know, mixing, mixing, you know, the different colorants. And so, the plaintiff in that case had invented a blue noise mask, but over, you know, several years of continuation practice basically claimed a half toning mask that basically I think that term was half toning mask designed to produce substantially all visually pleasing profiles. So going for something like a blue noise mask or something that basically covered everything. And so in that situation, it wasn't so much about whether or not the patent was valid, it was whether or not the original disclosure covered that and whether or not they could get the benefit of the filing date when they broadened it out beyond the blue noise mask. And then we have several other cases where, again, that are completely different where, you know, we had the set of cases, the Rendy cases, which ended up being three different litigations over eight different patterns. We won the first case, but continuations were filed beyond that. And the technology had to do with basically analyzing information as it's typed in, and then doing some other action beyond that. So what was disclosed in the patent was basically you type in, you start typing in an address, and it looks it up in your, in your address book to see like, does it exist? And if so, you know, the software would fill in the rest of the address, or it would, you know, ask you, if you do want to add the address to your address book, this was then broadened to include all sorts of things that you know, that we were doing that you can see word, a lot of you I'm sure I've used word and you type in Word, and it starts to fill in the rest of the sentence for you, or it offers a bunch of other additional actions like, maybe it looks like you have a tracking number on this, you want to look up this, do you want to look up this package tracking number. And so over time, what happened was, you know, as we disclosed a bunch of prior art, that prior art got washed through the patent office. And in fact, actually, the patentee narrowed their claims by putting in a lot more detail about about like, how all this analysis was done, none of which was disclosed. And obviously, it was targeted directly at us because they had gotten discovery about exactly what we're using. So it's not so much necessarily about Brett, it's about just morphing the claims completely separate from what you actually disclosed in your in your original disclosure.
Speaker 3 16:48
That's really helpful. I want to actually ask a follow up question to that, which I'll start with Natalie. But I'd love to hear from all of you, which is, you know, what Isabella is talking about is kind of a common approach in at least we experienced in the tech industry, which is, you know, having continuations that kind of are evolving as litigation and discovery is taking place. And we had an article from former PTO Commissioner Bob Stoll a few weeks back, saying, you know, what would be a common sense approach to address that kind of behavior is simply requiring the applicant to point to the specification when they're making these changes, or when they're filing that continuation 10 years out. And his his thought was, it's not going to cost too much. It's, it's pretty straightforward, and really applicants in the best position to be doing this. So Natalie, what do you think about that approach from Bob's stole? How would that work in practice for you as a prosecutor?
Natalie S. Richer 17:42
So it could work, it is some extra effort, right to go through the claim that you want and put into detail exactly where each limitation where you're going to find the exact mapping to it? You know, a lot of times continuations, there's a decision whether or not you want to file it or not. And it's a cost benefit analysis of how great are the claims can I get and that can turn on a variety of different things, you know, what happened during prosecution? How easy is this examiner? how lenient dizzy, you know, or her? And, um, you know, or how much is this going to cost? This is going to be really difficult. Am I going to get it quickly? Is this going to be valuable? Or am I going to end up narrowing this anyway. So, you know, all those things go into it. And I think a lot of times for continuations, we're able to push the boundaries a little bit because we know from the prior patent, how prosecution went and what we're likely to do. I think if you start putting in the detailed support, where each limitation come from, it does shift the cost benefit analysis a little bit and it also might shift the perspective of what you're likely to get, you know, some of the more lenient examiners now have an exact roadmap where, you know, maybe they can't be so lenient anymore, with the support.
Speaker 4 19:07
So just to add to that, all good points. I really liked Bob's Bob Dole suggestion here. Taking into account Natalie's question about cost benefit shifting. So two things and when it's a little bit like tough love in the sense that if the Office requires this, it's sort of good for the applicant. Yeah, the cost a little bit now. But look, it's gonna force you to do your diligence and make sure your claim isn't overbroad backed Isabella's comment right keeping the claims tethered to the application, applicants really should be doing this themselves as a matter of a best practice. And so if the office at least encourages them to do it, if not requires them. I think it does. Cause the rigor that will later benefit the applicant by making sure that the claim is well they'll just by makes it easier for the examiner then to examine the claim and hopefully issue it, it makes the claim more defensible in litigation and avoids the further tension. With section one, two up, the one of the thing I'll say is, though, and this may be coming back to agree with Natalie, that I did not expect that this should and I don't think Bob still would have suggested that this shifts the prime aphasia burden from the examiner having to make a rejection if there's a rejection to be made. So it's kind of a, you know, best practice disclosure. Look at you, you're making the claim amendment. So you should tell us where it's rooted in the specification, but not a basis of not like a shifting of burdens. Isabel, if
Speaker 5 20:45
I could, yeah, pop into that. I mean, as the person who's paying bills, for the proceeds to the outside lawyers, I actually, I would really appreciate it to actually have this requirement, we actually have this requirement in Europe. So this is one of the reasons that we had also stressed this to the PTO. And what we were finding is that, you know, we might get a claim issued in the US based on a preseason disclosure, and then have problems in Europe. And so, since we're gonna have to do it for Europe, anyway, module, just do it here. And I think in terms of the cost, I mean, you know, I don't know what everyone requires for their outside counsel. But, you know, when we file a continuation, we actually expect them to show us where it is in the spec anyway. So it seems like it's pretty minimal work to actually go through that effort. And I, when I hear a lot of complaints from, you know, particularly small inventors or small companies about their patents and how it's difficult to enforce them, I sometimes look at those. And I think, well, if the if the PTO had had these rules, your prosecutor would not have been able to get away with this, or, you know, maybe they were just laser or maybe they're trying to pull a pass on, who knows. But I think that a lot of small companies that don't have in house counsel, to police these things and make sure that their counsel are doing the right thing would benefit quite a bit, because then they would have much clearer patents, that would be easier to enforce. And I just think, like I said, I think this is, you know, we're all dealing with us in Europe anyway. So it seems like a minimal amount of effort to try. And the result is that we do have less continuation practice. In Europe. I mentioned the render case that we had earlier. When we when we litigated that in Europe, you know, the there was a patent was issued, but the way that they were trying to read it. The you know, the court there, the German court basically said, this is invalid for added matter, which is their version of written description. So it was a much quicker and smooth process to go through it in Europe than here in the United States. Yeah.
Speaker 3 22:44
All right. Well, let's shift gears from 112, a and written description and enablement to definiteness in 112. B. And for software claims, the definiteness question often hinges on whether the claims are interpreted under Section 112 F, which is what we've been talking a lot about today without really saying it too much. That's functional claiming. So I wanted to give a very brief history before the panelists will will talk a bit about this. So the case you want to be thinking about when it comes to functional claiming is all the way back in 1946, with Halliburton, and that is where the Supreme Court held that if you're using functional language at the point of novelty, the claims are invalid, because they failed to satisfy the requirements of definiteness and written description. And the reason this was the case is the court was just concerned that you were describing an invention by what it does, and not the physical characteristics. And we're going to preempt everything. And there's a quote from the Supreme Court on this slide that basically outlines their concern, which is that you could basically frighten future innovators away from doing that follow on work with, which I think the last panel really talked to a lot about which is that subsequent innovator who otherwise might not be innovating in that space, because of these, these broad functional claims. So as a response to Halliburton, you had Congress, partially overruling it in the form of the 1952 Act. And this is what created what was section 112 Six, and now it's 112 F. And I'll read it out loud, just we all have this as a baseline, an element and a claim for a combination may be expressed as a means or a step for performing a specified function without the recital of structure material or acts in support thereof. And such claims shall be construed to cover the corresponding structure material or acts described in the spec and equivalents thereof. So the way I read this, the way everyone sees this is, it's okay to use functional claiming after Halliburton, but the claim should be construed under 112 F and limited to what you're describing in the specification. And so now let's fast forward to 2008 where Are we talk about this in a software context, in the aristocrat case where this, the court here, fed circuit looked at what this all means in the context of software, and specifically, where someone is just you know, has a general purpose computer involved. And you can see, what the court is saying here is just disclosing a general purpose computer as your structure, it's not going to cut it, because the general purpose computer can be programmed to do that function in many different ways. And you need to limit it to the corresponding structure material, or x, which would be an algorithm in this case.
Unknown Speaker 25:40
So we will
Speaker 3 25:41
hopefully talk a little bit about that algorithm piece of it, because I think that is a real open question and in our field, but first, I want to talk about 112. F in general in the software space, and I'm going to turn to Natalie, first to just give a little bit of meat on the bones here. So Natalie, why are software claims often written in this functional language? And can you share some examples of language that would be interpreted under this provision?
Speaker 6 26:06
Absolutely. So a lot of times the written in functional language, because the function is the innovation, we're programming a server or a computer or something to do something that's unique. You could, your code can vary in many different ways. But it's the innovation is the function that you were able to get this computer to do. And so, I've, there's been, I think, maybe an uptick recently that I've noticed in 112, F, even in the office actions that we received, that won't even be a rejection, it'll just be a statement that I'm interpreting your claims under 112 F. And it's sort of up to you whether you want to respond to it or not. Some of the language might kind of surprise you as to what invokes 112 F and what doesn't. And I took some notes based on an interview I had with an examiner recently. So bear in mind, this is based on one person, but what they said was 112 F is invoked if you use the word system, Module Manager or unit, but what would pull it out of 112 f if you use the word server, interface, logic, processor, or controller. So just a lot of people those might sound very similar. So it's, you know, this is what causes the, the questions.
Speaker 3 27:32
Well, now I have to do said, well, so I want to know what your reaction is to, can you parse those lists?
Speaker 4 27:38
I can't really it's the same as that is, it reminds me of a term that a great Federal Circuit Judge graduated from NYU, use nonce words, right? These just seem like they're all like previous versions of nonce words. And I can tell you that I'd be able to say one should be interpreted into 112 F in the other.
Speaker 3 28:01
Well, I want to turn keep with you, Dave, and go back to your time as PTO director, when I mentioned at the outset that this was an area you were thinking a lot about. And at the time, you mentioned results oriented claims, and express it that was a concern. Can you tell everyone what results oriented claims means why that was a concern for you, especially for software?
Speaker 4 28:24
Yeah, well, this came after the area case came out in 2010. And so we were grappling at the time at the office with how to examine better, in view of the area had case in the developing law on what 12 And it was our belief, I think, collectively in the office that if we could get better at 112 examination, we could be helpful to the information technology industry, in particular, and ensuring that better constraining claims were granted. And so these results oriented claims, and they're very similar, similar to what's called single means claims are a particular problem, because they claim every solution to a problem rather than a specific solution to a problem to show. We were trying to merge that thought with Natalie's point about the fact that look software. Inventions are a lot about process. And they are really are a lot about, you know, finding a new thing you can do with a computer which is which is inherently a result. Right. And there are many ways to do the program, similar problem to what our life sciences colleagues are dealing with today, right in the biotech industry. And this comes up of course, back to the Amgen case. So we were trying to figure out how to find a reasonable spot between those two places. But especially I was especially focused on the so called single means claims are resulting in one because I thought that was an example where you're just claiming a problem, right? You know, I think it would be great to have a heavier than air flying machine. Well, that's not a solution to a problem. That's a statement of above, that's a result that you'd like to obtain. So that's why they call them in the office, frequently, the column results obtained claims or results achieved claims are results oriented, all to say, I felt then still feel that the office needs to be rejecting results obtained claims as inherently exceeding the scope of 112. And they are such a scourge if they issue out of the office, because then you've got a claim that covers all solutions to a problem and not any particular solution.
Speaker 3 30:46
So let's, let's build from that and turn to Isabella, because I know you're eager to talk about Williamson and give more details on the case, because that case, of course, was very important case on 112. F. So can you talk about how it changed, or at least purported to change the interpretation of claims under 112? F? Yeah.
Speaker 5 31:04
I mean, I think you have to build on Dave's point. I mean, I think that's right. I think the office is concerned about issuing these claims that are too broad. But there's, there's another issue that goes beyond that, which is, you know, sometimes I think the office thinks it's issuing an error claim, because they think it will be subject to 112, AF, or 112. Six, and then there's years of litigation about whether or not that's the case, and I think that is a problem as well. And so, so the when was the case is is, you know, I think an illustration of that, like I to me, you know, there were six years of litigation on this case. And and the the term that I think it's a lot of attention here is it was they were having a distributed learning system. That was basically standard computers that displayed the same, the same output, like in two different computers in two different places, right. So and they call it sort of a virtual classroom environment. And if you think about that, like that's, this is why there were so many defendants in the case was us there was Cisco with WebEx, there's tricks which go to meeting today would be, you know, zoom, whatever Google needs everything. Right. And and the claim was that general and I, I actually think that the the office actually intended to be construed under 112. But of course, the patentee didn't agree because they wanted a really, really broad claim. And so it went up to the Federal Circuit, you know, we wanted this district court went up to the Federal Circuit, the first Federal Circuit decision on a three judge panel, you know, disagreed that it was 112. F, although judge Reyna was in the dissent on that, it then went on, bonk, and then it reverse, and so and so it was determined, that claim was determined to be subject to 112 F or 112. Six. And, you know, from that, like, my experience was, you know, I just wish there was some way or that there was a requirement for the office to say whether or not it was interpreting under 112, six or 112 F or not. And, in fact, it would actually even be easier if the patentee were required to say that, right? Because why otherwise you're spending, you're having the exam or spend a lot of time thinking about does this patentee intend for this to be covered under 112 f6? Because there's many reasons in the in the software world that you would want it to be covered that way, because these are way of claiming. And so to me, like, that's the issue here is like, why do we have so much litigation overall, this does another interesting thing about that, that, you know, the pattern that was in the Williamson cases, there were also method claims, which eventually were invalidated and or one on one. And so we've had a lot of discussion today about in play between one on one on 112. And I think they all do slightly different things, depending upon how you've claimed your pattern. So, you know, one way or another, again, the method claim was basically about sort of distance learning using two computers, standard computers over the internet, to display, you know, similar information in two different places. So, you know, you know, at bottom, I think, what's happening also with the office, and it's interesting to hear what was said about the nonce words, you know, in my experience, actually, in our, in our prosecution, we are seeing rejections based upon whether or not you use a word like, you know, circuitry, or module or routine or server, as opposed to digging into what's after the nonce word, which is what really matters. You know, you can have a you can have a claim that just says search to do this or module that does that, if you have an structure in the function such that it is it is more akin to the actual algorithm as opposed to just claiming a function. And so, you know, there's subtleties there about like, what's the function and what's really more akin to an algorithm, but I do think a lot more clarity in this will be held For all the litigants involved, and I do also, again, I look at some of the patents that are that are prosecuted on behalf of some of these small companies. And I feel like their prosecutors probably did them a little bit of a disservice in the way that they claimed things. And so I think a little more rigor from the office would help help those panties.
Speaker 3 35:21
So just I'm going to put on the screen behind me to just a reminder to everyone that Williamson was supposed to remove the strong presumption. Oh, yeah. No, it was it was about it. You know, before it was pretty hard if it didn't say means to say it was a means plus function claim. This was just basically like, we're not gonna let you play games with your words. And these nonsense words. You know, the ones Natalie mentioned in other shouldn't be your ticket out? 112 F, I think. And I want to ask Dave, you know, do you see a shift since Williamson? And I guess just to put it bluntly, did the removing the strong presumption matter?
Speaker 4 36:00
I think it absolutely did. And I think Williamson was a great case. It recognize the reality of what Isabel was pointing to that going there, there are lots of different nonword. And what really counts I love your point is about what really counts is what comes after the means floor or the more the server or in an interface or logic or whatever word you so Williamson was great. And it has had a huge impact. So of course, the office put out new guidance and trained examiner's. And there is no like strong presumption against 112 AP being applied, just because the word means that magic word isn't used, actually found a law 360 article where someone analyzed court decisions for the three years before Williamson, and the three years after Williamson. And what they found was that in district court litigation in the three years following Williams, and that litigants were much more likely to raise 112 F issues for claims involving words other than means for and that the contingents prevailed in the district court 75% more often than they did before went with it. So that's pretty strong statement of the result. And then lastly, you know, I mentioned PTO guidance came out and examiner's now from everything I've seen are much more likely to least ask questions, if not issue 112 F guidance or or assert that they're going to interpret claims under 112. Back to Natalie's point, when I was there, we had a recommendation that we made internally, and it was never put into into the MPP. But I'd love to see it put into the MPP. Bridging put together here that examiner's should require applicants where there's any question about whether 112 F might apply. To state that your claim you wrote it, you tell us how you want us to examine it and how you want it interpreted. And there, as you said, as well, there's good reasons to interpret claims and 112. So it's not like it's always an admission against interest, right? Just make up your mind. It's your claim, and we'll do whatever you want with it. And I'd love to see the office move this? Well, I
Speaker 3 38:39
want to see what Natalie thinks about that. Because as the prosecutor you here who will be putting on the record, please, you know, interpret my claim under 112. F, would you would you take them up on that offer?
Speaker 6 38:52
I can't think of a scenario when I read ever asked to interpreted under 112 F. I wouldn't personally mind the requirements if they said you tell me do you want this under 112 F. I don't love the way it's being presented recently where it's just kind of snuck in there and it's easy to miss where it says I'm interpreting this Senator 112 F. And there's no requirement to respond to it. So I think just by default, if you overlook it, or if somebody overlooks it, then are you stuck with it? Possibly. So I don't love that I would be okay with being forced to give give some response.
Speaker 4 39:39
It does seem like a fair point. It's a little like putting a rocket in your pocket right for the examiner to just sneak in a statement that says, I'm going to interpret this underwater. Maybe it would be better to have like a very affirmative discussion about it.
Speaker 3 39:51
Yeah. Yeah. And I think that goes to clarity of the record too. Because yeah, it's here is how we're doing everything as the examiner is putting it for worth rather than sort of hidden in the rejection, I think I looked at some data from the PTO, they have something called a master review form, which is what they use to review a handful of applications for quality. And there's a piece of that that talks about this very question, which is whether 112 F was invoked or not. And it looks like 92% of the time, there's no indication one way or the other. So might be that middle ground, you're talking about Natalie, where it actually was, but it's not kind of explicitly being said, we're using 112. F, here is the structure we're bringing in from the specification. And here's how that's playing out in the invalidity, because I, it sounds to me like it is happening more frequently. But maybe it's not being done explicitly enough. So that could be that could be a good area to move, especially as we get into the definiteness question, which naturally comes from this. Before we move on, I did want to raise a recent fed Circuit case, it was the DI fan case, from 2022. And I don't think this went anywhere. But this was a case where the word code was being used in a software claim. And that's one that comes up a lot, and maybe in these lists of words, and it was being used, and it was found to recite sufficient structure to avoid 112. F. And I think that that brought about an amicus brief from a long list of academics, including Professor Dreyfus from NYU and Mark Lemley and, and others. And they were worried that the word code in place of actual structure was kind of doing what you weren't supposed to be able to do at post Williamson. And, you know, Williamson was, quote, a dead letter based on this case, I just are their reactions to this one, because I didn't follow this case too much at the time, but preparing for this panel, I saw it and thought, that seems that seems problematic.
Speaker 5 41:50
Well, I'm not familiar with the case. But I do think, you know, if you go back and did the law changes results, Williamson or not, I mean, I It's interesting, because Oh, Flack, like from the prosecution staff, like after we won the Wilmington case, it was our case, and they didn't realize like, behind this whole thing. And I was like, well, that law has always been there. In fact, we had, we had one, like many cases on this issue before. And so I think what it did do, and it's interesting, the data that that you pointed out, Dave, I think it made people more likely to bring up the defense. But like, I found that we we wanted a lot when we when we did bring it even before that. And so to me, again, I think I don't know what what followed after the word code in that situation. But, you know, I had another case, like the MIT abacus case, which is about sort of color matching colorimetric matching between, you know, a scanned document and a render document, like in a display. And, you know, there there are two different terms there at issue on the 112. F issue. One was color and selection mechanism, which was deemed subject to 112 af. And then the other one was aesthetic correction circuitry, which was not deemed to be under 112. F and judge Michelle, actually, Chief judgment shall at the time, dissented on the aesthetic correction circuitry, meaning he thought that it should be construed under 112, section 112 F at the time. But all the analysis was was about the words that followed. And so whether the word is server or module or code, the question is whether or not there's enough, in the words that follow to be more than barely functional. And so I don't know. Yeah, I mean, I don't know if you have the, the fair fellow followed from that. But that, because what Ben follows in that is often very akin to what would be, you know, if you claimed it as a function, and then you had to go to disclosure to look at what that function was, or what the disclosure was? Often it is, and that, you know, I think it surprises people like how bare bones, some of the disclosure is, in some of the software cases, it's not, you don't have to disclose code, right? You, you generally do it in a flowchart form or you do it in a, you know, you have these modules, you know, you have these, you know, standard things that perform these functions, right, but it is more of a in a flowchart form than a, you know, line by line code form
Speaker 3 44:14
is that's a perfect segue then to definiteness. Because once you're in 112 F territory, and you're looking at a software claim, construed under 112 F, then you are now looking okay, what is the structure because this is a software claim. And in aristocrat, we saw that the courts, the structure is the the algorithm, it's the, you know, the computer program to perform that specific function. And if the specification does not sufficiently describe the algorithm, then the claim is indefinite as an attempt to claim every way of achieving that function. And I think you'll see a range of answers in the case law in terms of how much detail is needed to supply sufficient algorithm. Some find you know, that flowchart is sufficient. And as prosecutors, we know, flowchart is often the way this is done, but these flowcharts can be pretty high level, others see more willing to look at what one of skill in the art would really need in order to decide if that algorithm is sufficient or not. So I think right now, it's I would say, in my view, it's a little bit murky. And I was hoping to turn to the panelists, maybe starting with you, Dave, on the state of the law in terms of sufficiency of an algorithm, the level of detail needed, and I'll throw in their AI and how that's going to make this even more complicated.
Speaker 4 45:32
Yeah. So, you know, what I see going on is, courts are turning for the definiteness requirement to requiring more than, you know, the mere structure recited in some kind of a rote passion. So the Aoyama case is one example where the courts when the DISCLOSE structure is a computer program to carry out an algorithm, right, basically, what as Bill was saying a minute ago, the DISCLOSE structure is not a general purpose computer, but rather a that special purpose computer program to perform the disclosed algorithm. So they're collapsing the disclosure on to the exact device that is specified that is set forth in the specification. And I actually don't have a problem with that one other, you know, example, I mean, that that can seem harsh, but I think it's better to have that more constrained view than it is to have a description that collapses on the claim language, which was what you were otherwise. So frequently see, one other example, advanced ground Information Systems, which was a 2016 Federal Circuit case, where they looked at this term symbol generator, and said that, that is a computer implemented means plus function limitation, and that the specification of the patent in suit do not disclose an operative algorithm for the, for the claim elements, reciting symbol generator. So the court, you know, wasn't willing to go and try and like, search around or make something up, they're really holding the patent D to disclose and describe in a definite way, what was set forth in the claim. So I see the general direction for software claims. Similarly here to being in a in a constraining more exacting mode than we've seen before.
Speaker 3 47:51
So how does that play out in prosecution? How are you describing things in detail enough that it's satisfying this algorithm requirement, under 112? B.
Speaker 6 48:01
So in drafting, we try to cover all the bases, right, I like to start with the claim itself, so that I know what I want to get. And then I go back and start drafting the spec, right? Describe it as generally and broadly as I can, and then drill down, describe, you know, the algorithm in detail, describe what's performing the algorithm, the computing environment, that it could be implemented in a few examples of how it could be implemented, you know, some other language so that way, in case I use the word system, and the claim thinking it was cool, and then learned later, it's not cool, I can fall back on processor or something like that. So it's good to, you know, my personal approach is that I like to put that all in there. And that way, there's options for later on, that if you do face these issues during prosecution, you can go back and look and say, Okay, I did drill down a little here, I can pull from this example, I can pull from this diagram. And luckily, the PTO was very lenient about letting you pull things from wherever you want, as opposed to maybe the EPO where it has to be very tightly in the same paragraph.
Speaker 3 49:11
And so just just a brief follow up on that. I mean, did that shift for you? After aristocrat saying, for software, it is a special purpose computer, it's programmed you did an algorithm? Or has it always kind of been on this continuum of you're trying to get as much detail as possible.
Speaker 6 49:28
It's always kind of been there. So I don't know that it changed much in the way of how the drafting is performed. I mean, maybe there's a little more thought into word choice and just double checking, you know, it's, it's always helpful to to just have a second set of eyes to look at it. Sometimes when the person drafting they're so deeply ingrained in it, and they spoke to the engineer who speaks at such a high level that, you know, they think they covered it and then another person comes along and says, You're not You're not really explaining this enough. So I think that's always helpful. Ah, but other than that, I don't think it changed drastically.
Speaker 3 50:03
But let's, before we run out of time, I did want to get to something which I think the last panel kind of teed up really nicely, which is, you have a claim that is drafted functionally. And 112. F is not being applied, it's not being interpreted under 112. F. So you have a, a functional claim in this, I don't know, in between space where, where you're not under 112. So you don't have this pulling in the structure from the spec. So I, I want to understand what the panelists think about that. Maybe starting with Isabella, when you have a claim that sort of sitting out there like that it the last panel, I think, and I don't want to put words in anyone's mouth, and they said, That's okay. It can be definite. Do you agree?
Speaker 5 50:50
I mean, it will obviously, it depends on on the circumstances, sometimes, you know, it may not, it may not be definitely you obviously have to look at the disclosure. Typically, if something is not falling under 112, F or 112, six, there is a lot of detail, at least what I find is that there is there is something like a little bit of a flowchart in there. And so, in that situation, you're generally looking me if you're the accused infringer, you're looking at, you know, usually opens up some non infringement defenses. And so then you're not as concerned about validity. But I do think you should still go back to the spec, because there are situations where the claims have morphed enough that even if it might be definite, should the disclosure be there, there are situations where there's no disclosure of that, in the in the specifications, that's, you know, similar to some of these other cases that we've had where the patent was invalidated. Or, you know, somehow the, the court then determined that the claim really was narrower. So that definitely that exists. And I do think, you know, despite some of the what's happening with Williamson, I do think the courts are still struggling with it's a little bit and it is, I would say harder for me to predict, claim interpretation in these situations, certainly than it is for me to predict now, come on 101. And I think if you look at the Federal Circuit statistics on this, as well. There is there I think there's more affirmance on one on one than there is on claim construction, which is, you know, effectively what this is. So, yeah.
Speaker 3 52:27
Dave, what do you think? I mean, I, I think some folks might say after Halliburton with with the creation of 112, six, you can't have functional claims anymore, unless they are interpreted under that. Do you agree? What do you think about this issue?
Speaker 4 52:43
Yeah, you know, I'm seeing this pretty similar to how Isabel is, there's, there seems to be, at least on the part of the court, some willingness to pronounce such claims to be valid, despite the fact that they straddle that line. So an example that's come up in a number of cases recently, this claiming approach of saying that a physical structural or computer element is configured to and so there was a case this Vizio case just last year, where the claim had an apparatus comprising a storage adapted to store one or more image frames, a processor adapted to obtain a first image frame, expand the first image frame, do some other things with the imagery, very functional, but each in each case, preceded by in one case, a storage adapted to it. And the other one, a processor adapted to the district court found that storage and processor were just black boxes for performing a function. I think I would have agreed with them on that the with the District Court on that, and took the claim down under 112. F, since there was no corespun corresponding structure in the specification, Federal Circuit turned it around, and and said that storage and processor were sufficiently structural, they did not trigger 112 out. So you can see, you know, there's a line everyone's trying to find, and some mines are on one side and some are on the other.
Speaker 3 54:25
That's Natalie, we'll just have to keep drafting for both both sides. And
Unknown Speaker 54:30
that's a smart tactic, I think.
Speaker 3 54:32
Well, I think we have a few minutes left, I want to give time for questions. So yes, please. Yeah, I
Judge Raymond Chen 54:40
got a question. I just want to say I was a strong supporter. Because I felt like, you know, the past one years patents in terms of their disclosure, a lot thinner than the patents that were issued and granted back in the 80s and 90s and Too often you would see patents, where you would read the description. And it would feel more like a dream machine where nobody really actually spent any time working on developing something, they just had a really nice dream of how a machine could do lots of great things. But now that we have this rule, I do feel like there have been times where I've been a little queasy. And an example would be what if the means plus function limitation, and it's undoubtedly a means plus function limitation under Williamson is for limitation that is not at the point of novelty of the claim. What if it's for something, let's just say super routine and conventional. And people's skill in the art would know it precisely the very basic set of operations that one would undertake to perform that function, that non novel function? Well, nevertheless, under current case law, if there is no algorithm, if there's no disclosure of that very, very, very well known set of operations and the spec, then that claim can go down under 112, six, 112. Two. And so I was just wondering, what does the panel think of that situation? And then a corollary situation is what if the limitation, the means plus function limitation really is at the point of novelty, this is the key moment in the claim. And yet, at the same time, you know, in your heart of hearts, that it's really not that hard to figure out, what would be an appropriate set of operations to perform that function? And then you see that that's not disclosed in the specification. Do you feel like, yeah, for policy reasons, or whatever else claim like that really needs to be put down? Thanks.
Speaker 5 57:08
So, you know, I think I mentioned before that we've made a suggestion to the to the PTO to have the patentee designate whether or not they thought 112, six should apply or not to a particular claim. But when I made that suggestion, it was only to novel like, at the point of novelty, so I do agree that often, you'll see claims, and I'll have like, you know, a storage device, you know, display device, all this other stuff, none of which is actually important to the claim, it's just the thing that attaches. But really, maybe the novelty is around, I don't know, some sort of compression algorithm or something like that. So I do agree that if it's not the point of novelty, you know, you should not you should not be faulted, if you disclose it in your spec is something like, you know, sort of standard storage device, such as XY and Z. Like, I think that's that should be sufficient disclosure in that situation. Since that's not the point of novelty. I do think in the point of novelty, the whole point of the patent system, is to have that point of novelty described in a way that other people can then build upon it, they can license it, it's to get the information out there. And so I think in that instance, the disclosure should be a little bit more robust. And the difficult thing, I think, in the computer science area, in particular, is that a lot of enablement is not really that difficult in a lot of situations. It's really, it's about, you know, how definite is this? And is this the best way of doing it in this situation? Right. And so I do think, if that is your point of novelty, I think you shouldn't be held to a higher standard, but where it's not I do agree like that you should be able to disclose, you know, whatever standard printer or standard display whatever. Other reactions.
Speaker 6 58:49
Yeah, I agree, it makes me queasy to to think that a non essential limitation sort of could bring a claim down under 112. F, because from a drafting perspective, if you had to list everything in the world, for every tiny thing you were describing, it would take forever, no, no client is going to want to pay you for 1000 page application where you're, you know, describing everything. So I certainly agree with that part. For the more substantive, you know, the point of novelty, you know, if it invokes 112 F, clearly then it invokes 112 F. I mean, you wrote the claim, right?
Speaker 4 59:29
Yeah, just to comment that gets us right back to the Halliburton case. Right, which was all about means plus function language used at the point.
Speaker 3 59:37
Yeah, that was the whole the whole thrust of that one. Well, we have one minute for one.
Unknown Speaker 59:44
So if you had, like, let's say, a claim, if you had like a claim, I've seen this mechanical where, you know, there's not really a screaming point and novelty, it's just like a whole bunch of different parts and you look at it and it's just really getting around the or artists pretty crowded art. And then along the way, it says something like a processor to talk to a phone with some information, that seems to be a very common thing now, like, you know, do something and then talk to an app on a phone, or a smart device or whatever. Does that invoke the process? When you say processor to talk to the smart device? Does that then bring in you need the algorithm for the processor? As it talks to the device? Does that count as sort of the general purpose computer needing an algorithm and all that?
Speaker 6 1:00:30
I think you need the algorithm but I wouldn't say it invokes 112. I would I mean, I use that language all the time, right processor and communication with, you know, a client side device and just there's the structure rights and some of that it's, it's a processor, you can list out every processor, but it's you do need the algorithm to explain what is happening in that communication.
Speaker 4 1:00:54
Going with Isabella's point about what counts is the word after processor. Your articulation Natalie rings in my ear is not invoking 112 F in this particular case.
Unknown Speaker 1:01:09
All right, well, thank you so much to the amazing panelists.
Announcer 1:01:12
The Engelberg Center Live! podcast is a production of the Engelberg Center on Innovation Law and Policy at NYU Law and is released under a Creative Commons Attribution 4.0 International license. Our theme music is by Jessica Batke and is licensed under a Creative Commons Attribution 4.0 International license